Contents

miaViz implements plotting function to work with TreeSummarizedExperiment and related objects in a context of microbiome analysis. For more general plotting function on SummarizedExperiment objects the scater package offers several options, such as plotColData, plotExpression and plotRowData.

1 Installation

To install miaViz, install BiocManager first, if it is not installed. Afterwards use the install function from BiocManager and load miaViz.

if (!requireNamespace("BiocManager", quietly = TRUE))
    install.packages("BiocManager")
BiocManager::install("miaViz")
#> Bioconductor version 3.18 (BiocManager 1.30.22), R 4.3.1 (2023-06-16)
#> Warning: package(s) not installed when version(s) same as or greater than current; use
#>   `force = TRUE` to re-install: 'miaViz'
#> Old packages: 'biomartr'
library(miaViz)
data(GlobalPatterns, package = "mia")

2 Abundance plotting

in contrast to other fields of sequencing based fields of research for which expression of genes is usually studied, microbiome research uses the more term Abundance to described the numeric data measured and analyzed. Technically, especially in context of SummarizedExperiment objects, there is no difference. Therefore plotExpression can be used to plot Abundance data.

plotAbundance can be used as well and as long as rank is set NULL, it behaves as plotExpression.

plotAbundance(GlobalPatterns, rank = NULL, 
              features = "549322", assay.type = "counts")

However, if the rank is set not NULL a bar plot is returned. At the same time the features argument can be set to NULL (default).

GlobalPatterns <- transformAssay(GlobalPatterns, method = "relabundance")
plotAbundance(GlobalPatterns, rank = "Kingdom", assay.type = "relabundance")

With subsetting to selected features the plot can be fine tuned.

prev_phylum <- getPrevalentTaxa(GlobalPatterns, rank = "Phylum",
                                detection = 0.01)
#> Warning: The 'getPrevalentTaxa' function is deprecated. Use
#> 'getPrevalentFeatures' instead.
plotAbundance(GlobalPatterns[rowData(GlobalPatterns)$Phylum %in% prev_phylum],
              rank = "Phylum",
              assay.type = "relabundance")

The features argument is reused for plotting data along the different samples. In the next example the SampleType is plotted along the samples. In this case the result is a list, which can combined using external tools, for example patchwork.

library(patchwork)
plots <- plotAbundance(GlobalPatterns[rowData(GlobalPatterns)$Phylum %in% prev_phylum],
                       features = "SampleType",
                       rank = "Phylum",
                       assay.type = "relabundance")
plots$abundance / plots$SampleType +
     plot_layout(heights = c(9, 1))

Further example about composition barplot can be found at Orchestrating Microbiome Analysis (Lahti, Shetty, and Ernst 2021).

3 Prevalence plotting

To visualize prevalence within the dataset, two functions are available, plotTaxaPrevalence, plotPrevalenceAbundance and plotPrevalence.

plotTaxaPrevalence produces a so-called landscape plot, which visualizes the prevalence of samples across abundance thresholds.

plotTaxaPrevalence(GlobalPatterns, rank = "Phylum",
                   detections = c(0, 0.001, 0.01, 0.1, 0.2))

plotPrevalenceAbundance plot the prevalence depending on the mean relative abundance on the chosen taxonomic level.

plotPrevalentAbundance(GlobalPatterns, rank = "Family",
                       colour_by = "Phylum") +
    scale_x_log10()

plotPrevalence plot the number of samples and their prevalence across different abundance thresholds. Abundance steps can be adjusted using the detections argument, whereas the analyzed prevalence steps is set using the prevalences argument.

plotPrevalence(GlobalPatterns,
               rank = "Phylum",
               detections = c(0.01, 0.1, 1, 2, 5, 10, 20)/100,
               prevalences = seq(0.1, 1, 0.1))

4 Tree plotting

The information stored in the rowTree can be directly plotted. However, sizes of stored trees have to be kept in mind and plotting of large trees rarely makes sense.

For this example we limit the information plotted to the top 100 taxa as judged by mean abundance on the genus level.

library(scater)
library(mia)
altExp(GlobalPatterns,"Genus") <- agglomerateByRank(GlobalPatterns,"Genus")
altExp(GlobalPatterns,"Genus") <- addPerFeatureQC(altExp(GlobalPatterns,"Genus"))
rowData(altExp(GlobalPatterns,"Genus"))$log_mean <-
    log(rowData(altExp(GlobalPatterns,"Genus"))$mean)
rowData(altExp(GlobalPatterns,"Genus"))$detected <- 
    rowData(altExp(GlobalPatterns,"Genus"))$detected / 100
top_taxa <- getTopTaxa(altExp(GlobalPatterns,"Genus"),
                       method="mean",
                       top=100L,
                       assay.type="counts")
#> Warning: The 'getTopTaxa' function is deprecated. Use 'getTopFeatures' instead.

Colour, size and shape of tree tips and nodes can be decorated based on data present in the SE object or by providing additional information via the other_fields argument. Note that currently information for nodes have to be provided via the other_fields arguments.

Data will be matched via the node or label argument depending on which was provided. label takes precedent.

plotRowTree(altExp(GlobalPatterns,"Genus")[top_taxa,],
            tip_colour_by = "log_mean",
            tip_size_by = "detected")
Tree plot using ggtree with tip labels decorated by mean abundance (colour) and prevalence (size)

Figure 1: Tree plot using ggtree with tip labels decorated by mean abundance (colour) and prevalence (size)

Tip and node labels can be shown as well. Setting show_label = TRUE shows the tip labels only …

plotRowTree(altExp(GlobalPatterns,"Genus")[top_taxa,],
            tip_colour_by = "log_mean",
            tip_size_by = "detected",
            show_label = TRUE)
Tree plot using ggtree with tip labels decorated by mean abundance (colour) and prevalence (size). Tip labels of the tree are shown as well.

Figure 2: Tree plot using ggtree with tip labels decorated by mean abundance (colour) and prevalence (size)
Tip labels of the tree are shown as well.

… whereas node labels can be selectively shown by providing a named logical vector to show_label.

Please note that currently ggtree can only plot node labels in a rectangular layout.

labels <- c("Genus:Providencia", "Genus:Morganella", "0.961.60")
plotRowTree(altExp(GlobalPatterns,"Genus")[top_taxa,],
            tip_colour_by = "log_mean",
            tip_size_by = "detected",
            show_label = labels,
            layout="rectangular")
Tree plot using ggtree with tip labels decorated by mean abundance (colour) and prevalence (size). Selected node and tip labels are shown.

Figure 3: Tree plot using ggtree with tip labels decorated by mean abundance (colour) and prevalence (size)
Selected node and tip labels are shown.

Information can also be visualized on the edges of the tree plot.

plotRowTree(altExp(GlobalPatterns,"Genus")[top_taxa,],
            edge_colour_by = "Phylum",
            tip_colour_by = "log_mean")
Tree plot using ggtree with tip labels decorated by mean abundance (colour) and edges labeled Kingdom (colour) and prevalence (size)

Figure 4: Tree plot using ggtree with tip labels decorated by mean abundance (colour) and edges labeled Kingdom (colour) and prevalence (size)

5 Graph plotting

Similar to tree data, graph data can also be plotted in conjunction with SummarizedExperiment objects. Since the graph data in itself cannot be stored in a specialized slot, a graph object can be provided separately or as an element from the metedata.

Here we load an example graph. As graph data, all objects types accepted by as_tbl_graph from the tidygraph package are supported.

data(col_graph)

In the following examples, the weight data is automatically generated from the graph data. The SummarizedExperiment provided is required to have overlapping rownames with the node names of the graph. Using this link the graph plot can incorporate data from the SummarizedExperiment.

plotColGraph(col_graph,
             altExp(GlobalPatterns,"Genus"),
             colour_by = "SampleType",
             edge_colour_by = "weight",
             edge_width_by = "weight",
             show_label = TRUE)
#> This graph was created by an old(er) igraph version.
#>   Call upgrade_graph() on it to use with the current igraph version
#>   For now we convert it on the fly...

As mentioned the graph data can be provided from the metadata of the SummarizedExperiment.

metadata(altExp(GlobalPatterns,"Genus"))$graph <- col_graph

This produces the same plot as shown above.

6 Plotting of serial data

if(!requireNamespace("devtools", quietly = TRUE)){
    BiocManager::install("devtools")
}
if(!requireNamespace("miaTime", quietly = TRUE)){
    devtools::install_github("microbiome/miaTime", upgrade = "never")
}
# Load data from miaTime package
library("miaTime")
data(SilvermanAGutData, package="miaTime")
tse <- SilvermanAGutData
tse <- transformAssay(tse, method = "relabundance")
taxa <- getTopTaxa(tse, 2)

Data from samples collected along time can be visualized using plotSeries. The x argument is used to reference data from the colData to use as descriptor for ordering the data. The y argument selects the feature to show. Since plotting a lot of features is not advised a maximum of 20 features can plotted at the same time.

plotSeries(tse,
           x = "DAY_ORDER",
           y = taxa,
           colour_by = "Family")

If replicated data is present, data is automatically used for calculation of the mean and sd and plotted as a range. Data from different assays can be used for plotting via the assay.type.

plotSeries(tse[taxa,],
           x = "DAY_ORDER",
           colour_by = "Family",
           linetype_by = "Phylum",
           assay.type = "relabundance")

Additional variables can be used to modify line type aesthetics.

plotSeries(tse,
           x = "DAY_ORDER",
           y = getTopTaxa(tse, 5),
           colour_by = "Family",
           linetype_by = "Phylum",
           assay.type = "counts")

7 Plotting factor data

To visualize the relative relations between two groupings among the factor data, two functions are available for the purpose; plotColTile and plotRowTile.

data(GlobalPatterns, package="mia")
se <- GlobalPatterns
plotColTile(se,"SampleType","Primer") +
  theme(axis.text.x.top = element_text(angle = 45, hjust = 0))

8 DMN fit plotting

Searching for groups that are similar to each other among the samples, could be approached with the Dirichlet Multinomial Mixtures (Holmes, Harris, and Quince 2012). After using runDMN from the mia package, several k values as a number of clusters are used to observe the best fit (see also getDMN and getBestDMNFit). To visualize the fit using e.g. “laplace” as a measure of goodness of fit:

data(dmn_se, package = "mia")
names(metadata(dmn_se))
#> [1] "DMN"
# plot the fit
plotDMNFit(dmn_se, type = "laplace")
#> Warning: 'getDMN' is deprecated.
#> Use 'cluster' instead.
#> See help("Deprecated") and help("Now runDMN and calculateDMN are deprecated. Use cluster with DMMParam parameter and full parameter set as true instead.-deprecated").

9 Serial data ordination and trajectories

Principal Coordinates Analysis using Bray-Curtis dissimilarity on the hitchip1006 dataset:

library(miaTime)
data(hitchip1006, package = "miaTime")
tse <- hitchip1006
tse <- transformAssay(tse, method = "relabundance")
## Ordination with PCoA with Bray-Curtis dissimilarity
tse <- runMDS(tse, FUN = vegan::vegdist, method = "bray", name = "PCoA_BC",
              assay.type = "relabundance", na.rm = TRUE)
# plot
p <- plotReducedDim(tse, dimred = "PCoA_BC")
p

Retrieving information about all available trajectories:

library(dplyr)

# List subjects with two time points
selected.subjects <- names(which(table(tse$subject)==2))

# Subjects counts per number of time points available in the data
table(table(tse$subject)) %>% as.data.frame() %>%
    rename(Timepoints=Var1, Subjects=Freq)

Lets look at all trajectories having two time points in the data:

# plot
p + geom_path(aes(x=X1, y=X2, group=subject), 
              arrow=arrow(length = unit(0.1, "inches")),
              # combining ordination data and metadata then selecting the subjects
              # Note, scuttle::makePerCellDF could also be used for the purpose.
              data = subset(data.frame(reducedDim(tse), colData(tse)),
                            subject %in% selected.subjects) %>% arrange(time))+
    labs(title = "All trajectories with two time points")+
    theme(plot.title = element_text(hjust = 0.5))

Filtering the two time point trajectories by divergence and displaying top 10%:

library(miaTime)
# calculating step wise divergence based on the microbial profiles
tse <- getStepwiseDivergence(tse, group = "subject", time_field = "time")
# retrieving the top 10% divergent subjects having two time points
top.selected.subjects <- subset(data.frame(reducedDim(tse), colData(tse)),
                            subject %in% selected.subjects) %>% 
    top_frac(0.1, time_divergence) %>% select(subject) %>% .[[1]]
# plot
p + geom_path(aes(x=X1, y=X2,
                  color=time_divergence, group=subject),
              # the data is sorted in descending order in terms of time
              # since geom_path will use the first occurring observation
              # to color the corresponding segment. Without the sorting
              # geom_path will pick up NA values (corresponding to initial time
              # points); breaking the example.
              data = subset(data.frame(reducedDim(tse), colData(tse)),
                            subject %in% top.selected.subjects) %>% 
                  arrange(desc(time)),
              # arrow end is reversed, due to the earlier sorting.
              arrow=arrow(length = unit(0.1, "inches"), ends = "first"))+
    labs(title = "Top 10%  divergent trajectories from time point one to two")+
    scale_color_gradient2(low="white", high="red")+
    theme(plot.title = element_text(hjust = 0.5))

Plotting an example of the trajectory with the maximum total divergence:

# Get subject with the maximum total divergence
selected.subject <- data.frame(reducedDim(tse), colData(tse)) %>%
    group_by(subject) %>% 
    summarise(total_divergence = sum(time_divergence, na.rm = TRUE)) %>%
    filter(total_divergence==max(total_divergence)) %>% select(subject) %>% .[[1]]
# plot
p +  geom_path(aes(x=X1, y=X2, group=subject),
              data = subset(data.frame(reducedDim(tse), colData(tse)),
                            subject %in% selected.subject) %>% arrange(time),
              arrow=arrow(length = unit(0.1, "inches")))+
    labs(title = "Longest trajectory by divergence")+
    theme(plot.title = element_text(hjust = 0.5))

More examples and materials are available at Orchestrating Microbiome Analysis (Lahti, Shetty, and Ernst 2021).

10 Session info

sessionInfo()
#> R version 4.3.1 (2023-06-16)
#> Platform: x86_64-pc-linux-gnu (64-bit)
#> Running under: Ubuntu 22.04.3 LTS
#> 
#> Matrix products: default
#> BLAS:   /home/biocbuild/bbs-3.18-bioc/R/lib/libRblas.so 
#> LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.10.0
#> 
#> locale:
#>  [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
#>  [3] LC_TIME=en_GB              LC_COLLATE=C              
#>  [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
#>  [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
#>  [9] LC_ADDRESS=C               LC_TELEPHONE=C            
#> [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       
#> 
#> time zone: America/New_York
#> tzcode source: system (glibc)
#> 
#> attached base packages:
#> [1] stats4    stats     graphics  grDevices utils     datasets  methods  
#> [8] base     
#> 
#> other attached packages:
#>  [1] scater_1.30.0                   scuttle_1.12.0                 
#>  [3] patchwork_1.1.3                 miaViz_1.10.0                  
#>  [5] ggraph_2.1.0                    ggplot2_3.4.4                  
#>  [7] mia_1.10.0                      MultiAssayExperiment_1.28.0    
#>  [9] TreeSummarizedExperiment_2.10.0 Biostrings_2.70.0              
#> [11] XVector_0.42.0                  SingleCellExperiment_1.24.0    
#> [13] SummarizedExperiment_1.32.0     Biobase_2.62.0                 
#> [15] GenomicRanges_1.54.0            GenomeInfoDb_1.38.0            
#> [17] IRanges_2.36.0                  S4Vectors_0.40.0               
#> [19] BiocGenerics_0.48.0             MatrixGenerics_1.14.0          
#> [21] matrixStats_1.0.0               BiocStyle_2.30.0               
#> 
#> loaded via a namespace (and not attached):
#>   [1] RColorBrewer_1.1-3          jsonlite_1.8.7             
#>   [3] magrittr_2.0.3              magick_2.8.1               
#>   [5] ggbeeswarm_0.7.2            farver_2.1.1               
#>   [7] rmarkdown_2.25              fs_1.6.3                   
#>   [9] zlibbioc_1.48.0             vctrs_0.6.4                
#>  [11] memoise_2.0.1               DelayedMatrixStats_1.24.0  
#>  [13] RCurl_1.98-1.12             ggtree_3.10.0              
#>  [15] htmltools_0.5.6.1           S4Arrays_1.2.0             
#>  [17] BiocNeighbors_1.20.0        SparseArray_1.2.0          
#>  [19] gridGraphics_0.5-1          sass_0.4.7                 
#>  [21] bslib_0.5.1                 plyr_1.8.9                 
#>  [23] DECIPHER_2.30.0             cachem_1.0.8               
#>  [25] igraph_1.5.1                lifecycle_1.0.3            
#>  [27] pkgconfig_2.0.3             rsvd_1.0.5                 
#>  [29] Matrix_1.6-1.1              R6_2.5.1                   
#>  [31] fastmap_1.1.1               GenomeInfoDbData_1.2.11    
#>  [33] digest_0.6.33               aplot_0.2.2                
#>  [35] colorspace_2.1-0            ggnewscale_0.4.9           
#>  [37] irlba_2.3.5.1               RSQLite_2.3.1              
#>  [39] vegan_2.6-4                 beachmat_2.18.0            
#>  [41] labeling_0.4.3              fansi_1.0.5                
#>  [43] polyclip_1.10-6             abind_1.4-5                
#>  [45] mgcv_1.9-0                  compiler_4.3.1             
#>  [47] bit64_4.0.5                 withr_2.5.1                
#>  [49] BiocParallel_1.36.0         viridis_0.6.4              
#>  [51] DBI_1.1.3                   ggforce_0.4.1              
#>  [53] MASS_7.3-60                 DelayedArray_0.28.0        
#>  [55] bluster_1.12.0              permute_0.9-7              
#>  [57] tools_4.3.1                 vipor_0.4.5                
#>  [59] beeswarm_0.4.0              ape_5.7-1                  
#>  [61] glue_1.6.2                  nlme_3.1-163               
#>  [63] grid_4.3.1                  cluster_2.1.4              
#>  [65] reshape2_1.4.4              generics_0.1.3             
#>  [67] gtable_0.3.4                tidyr_1.3.0                
#>  [69] BiocSingular_1.18.0         tidygraph_1.2.3            
#>  [71] ScaledMatrix_1.10.0         utf8_1.2.4                 
#>  [73] ggrepel_0.9.4               pillar_1.9.0               
#>  [75] stringr_1.5.0               yulab.utils_0.1.0          
#>  [77] splines_4.3.1               dplyr_1.1.3                
#>  [79] tweenr_2.0.2                treeio_1.26.0              
#>  [81] lattice_0.22-5              bit_4.0.5                  
#>  [83] tidyselect_1.2.0            DirichletMultinomial_1.44.0
#>  [85] knitr_1.44                  gridExtra_2.3              
#>  [87] bookdown_0.36               xfun_0.40                  
#>  [89] graphlayouts_1.0.1          stringi_1.7.12             
#>  [91] lazyeval_0.2.2              ggfun_0.1.3                
#>  [93] yaml_2.3.7                  evaluate_0.22              
#>  [95] codetools_0.2-19            tibble_3.2.1               
#>  [97] BiocManager_1.30.22         ggplotify_0.1.2            
#>  [99] cli_3.6.1                   munsell_0.5.0              
#> [101] jquerylib_0.1.4             Rcpp_1.0.11                
#> [103] parallel_4.3.1              blob_1.2.4                 
#> [105] sparseMatrixStats_1.14.0    bitops_1.0-7               
#> [107] decontam_1.22.0             viridisLite_0.4.2          
#> [109] tidytree_0.4.5              scales_1.2.1               
#> [111] purrr_1.0.2                 crayon_1.5.2               
#> [113] rlang_1.1.1                 cowplot_1.1.1

References

Holmes, Ian, Keith Harris, and Christopher Quince. 2012. “Dirichlet Multinomial Mixtures: Generative Models for Microbial Metagenomics.” PLOS ONE. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0030126.

Lahti, Leo, Sudarshan Shetty, and Felix GM Ernst. 2021. “Orchestrating Microbiome Analysis.” https://microbiome.github.io/OMA/.